A Clustering Approach to Detect Imposter Syndrome Among Sri Lankan Undergraduates

Imposter Syndrome is another name for perceived fraudulence, which is characterized by feelings of personal inadequacy and self-doubt that endure despite education, achievement, experience and success. This is not a disease or abnormality, so there is no obvious reason to imposter emotions. Therefore, even if they suffer from imposter syndrome, they are not able to know this. The results of an undergraduate with imposter syndrome may be inappropriate academic choices, the impact on mental health and social isolation. The aim of the present study is to develop a computerized framework based on a data mining strategy to identify the Severity Level of imposter syndrome for Sri Lankan undergraduates. Thus, this research shows whether the person suffers from imposter syndrome as Low or Moderate or High in level. During the model development, a formal questionnaire was developed examining different influencing factors like depression, anxiety, parentification, family expectations, perfectionism, and low trait self-esteem that can affect the imposter syndrome of an undergraduate and was used to collect data from Sri Lankan undergraduates. In this study, five different unsupervised machine learning techniques, namely K-means, K-medoids, Spectral Clustering, Hierarchical Clustering and Gaussian Mixture Model Clustering were used. Clustering was selected as the best approach as it allows to detect patterns and similarities associated with undergraduates linked to imposter syndrome. To calculate the goodness of the clustering algorithms, the Silhouette index and the Calinski-Harabasz index were used. Among these five clustering algorithms, the best result was shown in the three clusters of K-means Hence, the finalized method helps to predict and classify severity levels of imposter syndrome among Sri Lankan Undergraduates into three groups as low, moderate or high. The research found that among 316 data points, 32.28% showed a low level of imposter syndrome, 16.77% displayed a moderate level, and 50.95% exhibited a high level.